The NIH Roadmap Epigenome Consortium has generated comprehensive epigenome profiles for over 100 cell types and tissues. While these maps have provided many novel insights into the epigenetic processes and helped annotate the cis-regulatory elements in the human genome, in-depth analysis of these epigenome maps is confounded by the fact that each epigenome dataset actually contains a mixture of two haploid epigenomes, and there are substantial differences between the two. To address this problem we propose to incorporate haplotype information into the analysis of epigenomes of diverse human cell-types or tissues. We have developed a new strategy to reconstruct chromosome-span haplotypes by combining proximity-ligation and ultrahigh throughput sequencing, and have been able to reconstruct haploid genomes corresponding to 368 Roadmap epigenome datasets from over 25 cell-types or tissues. We will perform integrative analysis to achieve three specific aims: First, we will generate haplotype-resolved epigenomes for 25 cell-types and tissues from seven individuals, and identify genomic regions that demonstrate allelic bias in transcription, chromatin modification or DNA methylation in these tissues or cell types. Second, we will investigate allelic gene expression and long-range control mechanisms in human H1 ES cells and four ES-cell-derived early embryonic lineages, to determine whether allele-specific states of enhancers or promoters correspond to allelic transcription of target genes during differentiation of human ES cells. Third, we will investigate allelic gene expression and epigenetic modifications in diverse human tissues and cell types, and identify sequence variants that contribute to allelic gene expression. We expect these analyses will substantially advance our knowledge of epigenomic landscape and gene regulatory mechanisms in human cells.